What is the best way to measure the performance of a strategy focused on risk-adjusted return? David Blitz and Pim van Vliet answer this question in their article, Benchmarking Low-Volatility Strategies, published in the Journal of Index Investing.
Low-volatility strategies promise less risk and better risk-adjusted returns than a passive investment in market-capitalization-weighted index. With such a goal, it makes sense that evaluating the performance of these strategies must involve something more than simply comparing returns with a market-cap weighted index.
In their paper, Benchmarking Low-Volatility Strategies, David Blitz, Head, Robeco Quantitative Equity Research, and Pim van Vliet, Portfolio Manager, Robeco Quantitative Equity study the alternatives and recommend two solutions depending on how investors define risk.
"Selecting a benchmark for a low-volatility strategy is not as straightforward as it might seem," says Blitz. "To begin with, a benchmark should represent an investable portfolio." The problem is that a minimum-volatility portfolio can only be determined with hindsight. It is not, as the researchers say, "observable" beforehand.
Blitz and Van Vliet also ruled out choosing a benchmark based on investable minimum-volatility portfolios as described in academic literature. “Although the long-term volatility reduction achieved by these various approaches is similar, the differences in portfolio composition may result in return divergences in the short run,” says Van Vliet. “It is therefore ambiguous to elevate the status of any one particular approach to the level of benchmark for low-volatility investment strategies in general.”
Of course, there are also the MSCI Minimum Volatility (MV) indices that were introduced in 2008. While there are some factors in favor of using these indices as benchmarks, such as their rules-based construction, modest turnover and ability to be tracked at low cost by passive managers, the authors conclude that they are ultimately less than suitable as benchmarks. This is based on a lack of transparency and subjective decision-making involved in the composition of the indices.
Blitz and Van Vliet also note that the MSCI MV indexes are no more effective in reducing volatility than other, more simple approaches, such as ranking stocks on past beta or volatility.
“The composition of the MSCI MV indices is based on a complex optimization algorithm which involves a sophisticated proprietary risk model,” says Blitz. "A large number of subjective assumptions are involved, for example, related to rebalancing frequency, constraints on stock, country and sector weights as well as constraints on exposures to MSCI Barra risk indices and turnover.”
Given the number of subjective assumptions and lack of transparency, Blitz and Van Vliet believe the MSCI indices are closer to being active, not passive, investment strategies. “A benchmark should be passive," says Blitz. "And with a passive strategy, you would expect little or no subjectivity to be involved.”
Blitz and Van Vliet conclude that the best approach for investors is to benchmark low-volatility managers against a capitalization-weighted market index and to adjust for risk.
“A straight comparison of returns is not appropriate given that low-volatility strategies tend to exhibit significantly lower risk in terms of volatility and beta,” notes Blitz. He suggests comparing Sharpe ratios, which measure risk-adjusted returns, or Jensen’s alpha, which evaluates the average return of a portfolio above what is predicted by the capital asset pricing model, based on the portfolio's beta and the average market return.
“The choice depends on how the investor defines risk,” notes Van Vliet. “If total volatility, meaning both systematic and idiosyncratic volatility, is most important, the Sharpe ratio is best. On the other hand, if systemic risk (beta) is more relevant, then Jensen’s alpha should be used instead.”
Van Vliet notes that the Sharpe ratio tends to be a more conservative measure, but may be difficult to interpret if returns are negative. Blitz adds, “When using our proposed measures, it may happen that all low-volatility managers turn out to have added value compared to the market index. In this case, it is still possible to distinguish between managers who provided more or less added value by directly comparing their Sharpe ratios or Jensen alphas.”
For more information, Benchmarking Low-Volatility Strategies.